Creating swarms of drones or robots is already complex. But underwater, even more so, since due to the environment, radio transmissions and GPS should not be relied on to guide machines and allow them to position themselves. However, three researchers from Harvard University (United States), specialists in engineering and biomimicry, managed to synchronize, in a 90 cm deep pool, seven robot-fishes.
The machines, whose shape is inspired by surgeonfish, were able not only to swim in schools but also to disperse under certain criteria before assembling and swimming together in a circle. All without Wi-Fi, Bluetooth and geolocation. The team relied on cameras embedded in the robots, so that they automatically position themselves according to what they see of the behavior of their fellows. A reproduction, in short, of what happens in nature. The project is detailed in an article published in Science Robotics for the month of January 2021.
Called Bluebot, these robots have four fins each and two 195 ° 3D cameras, both placed at the location of the eyes (there is only a 5 ° blind spot in the back of the robot, at the level of the caudal fin). They owe their name (blue robot) to three blue LEDs attached to their backs and stomachs. Underwater, the cameras of some capture the lights of other Bluebots. The size of the lights in a robot’s field of view gives an indication of how far away the LEDs are, and since every machine has more than one, an on-board algorithm can calculate their angular position.
–
No leader robot
A Bluebot can “see” up to 5m around it. However, there are still some limitations. In addition to the blind angle of 5 °, a robot cannot for example detect one which would be masked by a third.
In this system, there is no leading robot or centralized control of any kind. Everything takes place locally and by virtue of behaviors relating to each other. The researchers have in any case demonstrated the efficiency of this system but also the extreme maneuverability of the robots, capable of spinning at 150 mm per second at cruising speed and of diving at 75 mm per second or even operating a half. turn in less than 5 seconds. As the demonstration video reveals, when a robot is added to the bench that is already swimming in the pool, it synchronizes with the group immediately.
In this system, there is no leading robot or centralized control of any kind. Everything takes place locally and by virtue of behaviors relating to each other. The researchers have in any case demonstrated the efficiency of this system but also the extreme maneuverability of the robots, capable of spinning at 150 mm per second at cruising speed and of diving at 75 mm per second or even operating a half. turn in less than 5 seconds. As the demonstration video reveals, when a robot is added to the bench that is already swimming in the pool, it synchronizes with the group immediately.
It is possible to program swarms of robots to complete a mission without outside control
In addition to moving in a bench, the researchers also succeeded in developing a so-called search maneuver. A red light flashes in a corner of the pond and the fish are programmed to spot it (always via their cameras) and gather there. The first to succeed then opts for a stationary position while its blue LEDs start to flash. It is a signal that alerts the other robots, which stop their “search” and converge on the blue flashing.
The experiment therefore succeeded in combining several operations (movement, assembly, detection) into one. And the project tends to show that it is possible to program swarms of robots to fulfill a mission without remote control, without external control. Detection, observation and environmental monitoring missions, for example.
–
